16 Jul
2018
16 Jul
'18
6:19 p.m.
On Jul 16, 2018, at 10:22 AM, Clem Cole <clemc at
ccc.com> wrote:
...
We were were discussing at lunch one day, I don't think DEC thought there just was
much of a market, particularly once Pascal showed up. From the same discussion, one of the
compiler implementors said (remember DEC charged for its compilers so renumeration was an
important factor when they decided to build one - so it had to make money of they were to
build it): "I don?t recall any serious discussion of an internal DEC Algol compiler
project on the PDP-11, VAX or Alpha at DEC. To the best of my recollection, the only
serious users of Algol were British, and the language is a pain in the neck to implement,
with lots of performance land-mines."
Program managers have to estimate the business case for anything they do, and it's an
imperfect science. But still, those statements are not signs of careful analysis.
"Lots of performance land-mines"? Not so. Call by name is a simple function
call. Sure, if you use it when you don't need it your code slows down, but that's
no different from the performance hit you get in C++ if you put "virtual" on
functions that don't need it. In general ALGOL is no harder than PASCAL.
As for "only serious users" -- it's not clear to me whether that's being
said about ALGOL 68 or ALGOL 60. If 68, then maybe, though CDC found enough business for
it to sell the compiler (which was developed in Holland). If ALGOL 60, that's a
different thing entirely; it was widely used in Europe starting around 1962, and of course
it's also the implementation language of the Burroughs 5500 mainframe system. For
that matter, CDC, IBM, and probably others sold ALGOL 60 compilers.
paul